Alaina's profile - activity

Confirmed, a sudo dpkg -r yaml-cpp fixed the problem for me, which was trying to install Indigo on a netbook I upgraded from 12.04 to 14.04. (I also had to do a sudo dpkg -r sdformat for a similar error.)

We bought an Asus 1215N Eee PC with the intention of rebuilding our Turtlebot on it. Bizarrely, I can't get the default build of Ubuntu 10.04 to work correctly; in particular, the Broadcom wireless NIC and the Atheros Fast Ethernet interface are not supported. Ubuntu (including the kernel) are now up to date with the software update.

After fighting with it, I used a USB stick and my Mac to download the common-wireless drivers. I can build for the Atheros and Broadcom devices, but I can only make one of them work at a time. Once I got FastE working with common-wireless, I tried to install the default Broadcom hardware drivers... that didn't work either. The driver is active but I am unable to connect to any wi-fi network.

Isn't the 1215N the exact machine that Willow Garage is distributing? If so, does anybody know if a) Asus changed the hardware; or b) there's a simple way to get both network interfaces working at the same time?

Hi folks,

We are building our own TurtleBot and I am having problems with the IMU.

I am using the SparkFun ADXL203/ADXRS613 IMU (SEN-09127, http://bit.ly/jw6iUL ).

What I am experiencing when I try to calibrate it is similar to the problem described at: http://answers.ros.org/question/1333/...

[INFO] [WallTime: 1308820084.864965] Estimating imu drift
[INFO] [WallTime: 1308820085.165909] Still waiting for imu
[INFO] [WallTime: 1308820085.466922] Still waiting for imu
[INFO] [WallTime: 1308820085.767989] Still waiting for imu

...but I haven't remapped anything.

The IMU is wired to +5V, ground and AIN on the Create through the pinouts on an Element Direct BAM ( http://bit.ly/iHaTB8 ). The BAM isn't the problem; AIN passes through to pin 4, just like it should. When I first saw the problem I wired the IMU up to +5V on a bench supply and slapped a multimeter between RATE and ground. When I move the IMU around, I can see the voltage change, so the IMU is working.

If I look at /imu/raw and move the Create around manually, the X/Y/Z values remain at 0 (I don't have the accelerometer pins connected to anything, FYI), so I know that the problem is somewhere between the IMU and the ROS gyro stuff.

I know that my next troubleshooting activity is to see if the Create is actually receiving what the IMU is giving it, but I don't really know the best way to do that, and in any case my gut tells me that this is a software or configuration (not hardware) problem.

Does anybody have any suggestions?

Alaina - Singularity University - Mountain View, CA

I was having the same problem. Check this thread to see if it matches what you're seeing. It fixed the problem for me.

http://answers.ros.org/question/2993/registered-depth-only-works-sometimes-with

The /sbin has already been fixed in trunk, and will soon be in debs (we had some delays due to spinning up for Oneric). I just added the mkdir -p, thanks.

In the future, things that are clearly a bug/defect are best posted as a ticket on the appropriate Trac instance (for turtlebot, that is https://kforge.ros.org/turtlebot/trac)

More news:

Even with the update, I'm still seeing curious behavior. The first time I ran openni_launch, it worked. Then I ran the turtlebot kinect.launch and it worked. After the Turtlebot version had been running for a few minutes, I got messages like:

[ WARN] [1322144717.802973318]: Camera Info for rgb stream has subscribers, but stream has stopped.
[ WARN] [1322148935.064305326]: Camera Info for rgb stream has subscribers, but stream has stopped.
[ WARN] [1322148966.416502492]: Camera Info for rgb stream has subscribers, but stream has stopped.
[ WARN] [1322149924.350446490]: Camera Info for rgb stream has subscribers, but stream has stopped.

A rostopic hz running on the rgb camera_info starts showing "no new messages", it disappeared from rviz, and nothing happened any more. At that point, the turtlebot kinect.launch stopped working and hasn't worked since. Disconnecting and reconnecting the Kinect's USB to the netbook doesn't fix it, and neither does a reboot.

But, I can roslaunch openni_launch openni_launch and it works for a while. If it stops publishing, I can just cancel and relaunch it.

Hope this helps.

By the way, my load average was running above 4 before I did the update. I thought it was my crappy netbook, so I didn't give it much thought, but it turns out that it's running at 0.3 post-update.

You might have a look at this answer by Patrick Mihelich to another similar kinect related question. I dont know if the new binaries are already out, but i doubt it. Maybe you could try to build the most recent version from source like suggested and give feedback to Patrick if it helped.

I was fighting with this issue for weeks with natty and electric and for me a clean reinstall and switch vom 32 to 64 bit helped a lot, but didnt solve the problem completely.

Patrick's work, as linked in his answer above, fixed the problem.

I can't get rviz to run reliably in a Parallels or VirtualBox VM. It definitely seems to be Ogre-related. I think you might be out of luck. :(

Daniel, thanks! I did exactly that (deleted everything in ~/.rviz) and it worked.

here are good instructions for adding a sparkfun gyro: http://answers.ros.org/question/717/is-an-imu-required-to-run-turtlebot-stack

That IMU you showed on sparkfun looks like it only has outputs for I2C. Are you sure you listed the right one? That one doesn't have a RATE pin.

The best way to verify operation is to trace it every step of the way. You said you can see the voltage change. Is this change appropriate for the sensor and your expectation? Does the voltage go up when rotating one way and down when another?

Is there some sort of application that you can use to verify what the BAM reports? Or get some sort of printout?

I am pretty excited to have Turtlebot on Electric! I'm keen to try it on 11.10 at some point, but I'm still using Natty for now.

I've just done the upgrade to ros-electric-turtlebot (finally!) and had an upstart problem along the way. In install.bash, the chmods are pointed at /usr/bin (should be /usr/sbin), and the mkdirs error out if you've already run the script once or otherwise created the stuff in /etc/ros.

There are easy fixes for both of these: point the chmods at /usr/sbin, and add -p to the mkdir calls.

Here's my install.bash. Feel free to use it, of course, if it correctly fixes the problems:

#!/bin/bash

# Run this as root, from the directory containing it!
#
# USAGE: sudo ./install.bash
#  or
# sudo ./install.bash usb0
#
# where usb0 is whatever network interface you want to set the robot
# up for.  wlan0 is the default.

interface=wlan0

if [ $# -gt 0 ]; then
    if [ "$1" != "" ]; then
        interface=$1
    fi
fi

echo "Installing using network interface $interface."

sed "s/wlan0/$interface/g" < turtlebot-start > /usr/sbin/turtlebot-start
chmod +x /usr/sbin/turtlebot-start
sed "s/wlan0/$interface/g" < turtlebot-stop > /usr/sbin/turtlebot-stop
chmod +x /usr/sbin/turtlebot-stop
sed "s/wlan0/$interface/g" < turtlebot.conf > /etc/init/turtlebot.conf

# Copy files into /etc/ros/electric/turtlebot
mkdir -p /etc/ros
mkdir -p /etc/ros/electric
cat turtlebot.launch > /etc/ros/electric/turtlebot.launch

echo '. /opt/ros/electric/setup.bash; export ROS_PACKAGE_PATH=/home/turtlebot/ros:${ROS_PACKAGE_PATH}' > /etc/ros/setup.bash